Jump to ContentJump to Main Navigation
Democratic Accountability in Latin America$

Scott Mainwaring and Christoper Welna

Print publication date: 2003

Print ISBN-13: 9780199256372

Published to Oxford Scholarship Online: April 2005

DOI: 10.1093/0199256373.001.0001

The Accountability Deficit in Latin America

Chapter:
(p.79) 4 The Accountability Deficit in Latin America
Source:
Democratic Accountability in Latin America
Author(s):

Brian F. Crisp

Matthew Soberg Shugart (Contributor Webpage)

Publisher:
Oxford University Press
DOI:10.1093/0199256373.003.0004

Abstract and Keywords

This chapter presents accountability as an inherently vertical relationship. It discusses the conflicts between vertical accountability, horizontal exchange, and superintendence in presidential systems. It is argued that the relations between agents of the state are a function of their connections to the ultimate principal — the citizenry. The proliferation of superintendence agencies may raise expectations that cannot be met without an overhaul of the institutions of vertical accountability in Latin America.

Keywords:   horizontal accountability, vertical accountability, presidential systems, accountability deficit, Latin America, horizontal exchange, state agencies

As a result of much human sacrifice, democracy made a comeback in Latin America in the 1980s and 1990s, and scholars turned their attention to the explanations for these transitions from authoritarianism. More than a decade has passed since most of the transitions and now the primary focus of most research on Latin American democracy is on the illusive concept of its ‘quality’. Even casual observation was sufficient to recognize that the re-establishment of elections guaranteed neither decision-makers responsive to popular will nor unfettered political rights and civil liberties. Unpopular economic policies, striking levels of corruption, recurring constitutional crises, ongoing civil unrest (often violent), and popular sympathy for coup attempts highlighted the level of disjunction between politicians and their citizenry.

Guillermo O'Donnell (1994) characterized these less than satisfying regimes as ‘delegative democracies’—as opposed to more liberal ‘representative democracies’. While evocative and intuitive, the term proved complex, defying parsimonious definition or systematic operationalization. In an effort to clarify the exact nature of the shortcomings, more recently O'Donnell (1999) has argued that liberal, representative relationships fail (or delegative democracy results) because there is a lack of ‘horizontal accountability’, defined as:

the existence of state agencies that are legally enabled and empowered and factually willing and able to take actions that span from routine oversight to minimal sanctions or impeachment in relation to actions or omissions by other agents or agencies of the state that may be qualified as unlawful (38).

(p.80) The concept of ‘horizontal accountability’ has spawned a considerable commentary, redefinition, and criticism—some of which is on display in the various chapters of this volume. We shall argue below that accountability is best understood as a concept that refers not only to ‘answerability’ for one's actions (see Mainwaring, this volume; Dunn 1999), but also to sanctions. In other words, holding someone accountable means not only learning what someone has done, but punishing him for it if what he has done is in some sense ‘wrong’. In this sense, our understanding of accountability differs from that of O'Donnell and Kenney in their chapters in this volume. In light of this understanding of what accountability is, then, we consider the very idea of horizontal accountability to be an oxymoron. The term, ‘horizontal’, when applied to a political system implies two or more actors or institutions at the same level, but if one considers accountability to involve the possibility of sanctions, then the ability of one actor to sanction another implies that the one who may mete out sanctions has authority over the other. In other words, accountability implies that the actors in a political relationship are not at the same level.

In an effort to have as concise a concept as possible, we limit the term, accountability, to the sanctioning rights that are inherent in hierarchical relationships, which by their very nature exist when actors are in a vertical relationship with one another—one is higher in the chain of authority than another and thus can sanction the other.1 Horizontal relationships, on the other hand, imply the checks and balances that exist between separate agencies that are sheltered from sanctioning by one another in order to maintain their (relative) independence. We argue that, in presidential democracies, the separate origin and survival of the executive and legislature makes them agents of the voters, not one another, and therefore not institutionally accountable to one another (Shugart and Carey 1992). We call relations between them horizontal exchange, in the sense that they are co-equal branches (hence in a horizontal relationship to one another) that must trade (i.e., exchange) with one another to produce policy. In the theory of separation of powers, first advanced by James Madison, these exchanges prevent a tyranny from developing, thereby protecting the rights of citizens.

(p.81) However, actual practice of democratic systems, especially in Latin America, frequently departs from this Madisonian theoretical expectation. Recognizing the limits of horizontal exchanges between (supposedly) independent legislative, executive, and judicial2 branches, designers of presidential constitutions have increasingly turned to new horizontal relationships within the state. Examples include public prosecutors, ombudsmen, and controllers-general that are endowed with some institutional independence—meaning they cannot be dismissed by the president or a simple legislative majority. We call these forms of horizontal relationships superintendence. This term captures their role as watchdogs and providers of information—a role that suggests they may fulfil the answerability component of accountability. However, if accountability is understood as sanctioning authority, few of these entities meet the standard for being agencies of accountability—‘horizontal’ or otherwise. Thus, for example, we do not consider the ombudsman's provision of information regarding human rights abuses committed by officials of the executive branch (including the police or armed forces) to be a relationship of accountability, for there may be no consequences of the revelation of the information. If there are consequences, it is because perhaps the president fires the subordinate employees who perpetrated the abuses, or the legislature, using its power of the purse, reduces or revokes funding for the office in question, or the voters refuse to re-elect a president deemed to be complicit in rights violations. Only with sanctioning has the actor been held accountable, but if the actor who reveals the information does not impose the sanctions, then that actor is not holding the abusers accountable. It is, however, providing crucial information for others to hold the abusers accountable. Such is the essence of the horizontal relationships that we define as superintendence in this chapter.

Decades of reform across Latin America have produced a bewildering array of non-elected agencies responsible for superintendence functions over the branches of government. The core of our argument is that these agencies have emerged as antidotes to the (p.82) perceived shortcomings of democratic accountability in Latin America, but that the roots of those shortcomings are not to be found in the horizontal dimension of the state. That is, the deficit of accountability lies in faulty vertical accountability—legislators who do not represent the values and preferences of the broad citizenry. That is, if the accountability of elected officials were working as intended—such that voters could and would punish misdeeds—separate agencies of superintendence would be unnecessary, at least from the theoretical tradition inspired by Madison and on which the very structure of presidential democracy is based. The proliferation of entities of superintendence in Latin America, then, must be seen as largely a product of discontent with the functioning of accountability and it represents an effort to find a way around the problem without tackling the roots of the accountability deficit. Accordingly, we shall review the deficit of vertical accountability, and suggest remedies that, we believe, would be more helpful in the long run to solving the accountability deficit.

The structure of our argument is as follows. First we elaborate on the notion of accountability as an inherently vertical relationship. Next we show how the constitutional design of accountability varies between parliamentary and presidential systems. We then elaborate on the difficult interplay between vertical accountability, horizontal exchange, and superintendence in presidential systems. Next we provide a descriptive overview of the institutions of the justice system in Latin America, assessing the degrees to which they are independent of the elected branches. Then we elaborate on how to strengthen vertical accountability through candidate selection procedures and electoral rules. Finally, we conclude by reiterating that relations among the agents of the state are a function of their connections to the ultimate principal—the citizenry—and note that the proliferation of superintendence agencies is likely to raise expectations that cannot be met without an overhaul of the institutions of vertical accountability in Latin America.

Accountability in Presidential Systems

We tackle the concept of accountability from the perspective of a principal–agent approach. The reason for doing so is that we are persuaded by the arguments of those scholars who see the modern principal–agent approach, as applied to politics, as most consistent with the theory of government advanced by James Madison (p.83) (Grofman 1987, 1989; Kiewiet and McCubbins 1991; see also Palmer 1995; Elster 1999; Laver and Shepsle 1999, Strøm 2000). The Federalist Papers explained Madison's theory, justifying the design of the original presidential system, that enshrined in the American constitution of 1789. From this perspective, decision-making in large entities, such as democratic states, implies delegation of authority. In agency relationships, the right to make a decision is assigned by a ‘principal’ to an ‘agent’, but this assignment, i.e. delegation, is conditional. That is, it continues only at the pleasure of the principal. That it may be withdrawn is the very essence of accountability. Only when the right to make a decision is subject to withdrawal can we understand a relationship founded on accountability to be in place. Thus delegation occurs within hierarchies when one person or entity, as agent, receives conditional authority from another person or entity, as principal. Relations of delegation run in one direction—from principal to agent—while relations of accountability run in the opposite direction—from agent to principal.

Accountability means that the principal has the right to withdraw the conditionally delegated authority altogether. This usually means dismissing (firing) the agent. However, especially in politics, often it means something short of immediate dismissal. It may mean refusing to renew a delegation relationship that has a fixed endpoint or it may mean simply downgrading the agent's authority, but allowing the agent to retain her office. For instance, voters as principals can exercise accountability over their legislator-agents only at election time. As elections may occur only at intervals of anywhere from two to six years,3 there are obviously ample opportunities for the agent to ‘shirk’ at the expense of the principal. This is one of the reasons that delegation relationships are more imperfect in democratic politics than in, for example, the firm.4 Holding an agent accountable may also imply—instead of removing the agent—withdrawing some portion of the delegated (p.84) authority. This form of accountability is especially common in presidential systems. For example, a legislature can rescind or allow to expire delegated decree authority or it can reinstate constitutional guarantees, the suspension of which gave the executive ‘emergency’ powers.

One of our key points is that when institutions are formally independent of one another—as in presidential systems—they are not accountable to one another. Independence and accountability are two contradictory features of institutional design. The legislative majority in a prototypical presidential democracy cannot dismiss the president and only the voters (or not even the voters) can decide whether or not to renew the president's right to exercise the executive authority for a new term.5 While one branch may be answerable to another, this answerability takes place within the broader context of the ultimate accountability of all politicians to citizen-principals. That is why the way in which delegation and accountability relationships between citizens and politicians are structured is so crucial to the overall functioning of accountability—a theme we return to at length in a later section of this chapter.

While the executive branch is not an agent of the legislature in a presidential system in the sense of deriving its authority from the legislature, the executive can be the legislature's agent in the carrying out of discrete tasks. For instance, Congress may delegate by statute the right to negotiate trade agreements to the president, as under the United States ‘fast track’ trade negotiating authority, which existed until the Congress let it expire in 1995. In delegating such authority, it establishes a ‘contract’ that states the terms under which the president may bargain with foreign governments and the terms under which an agreement may be brought before the Congress for ratification. In this sense, the president is the Congress's agent for the purposes of forging trade agreements and is accountable to the Congress for the conduct of his trade policy and for the ultimate enactment of any trade agreement he (p.85) negotiates. If the Congress is unsatisfied with the president's actions in the area of trade, it can withdraw the delegated authority, either by passing legislation rescinding the right to negotiate future trade agreements or by failing to renew that right when it comes up for ‘sunset’ review. Note that in this example, the principal is not denying the agent his position (through dismissal). The president remains president, but with somewhat diminished authority.

In the various examples just cited, the agency relationship is vertical in the sense that the principal is holding accountable an agent who is, by definition, inferior to the principal in a hierarchy. The electorate is superior to the legislator in the hierarchy of democracy, and can hold the legislator accountable by not renewing her authority to exercise delegated rights as the voters' representative. In trade policy—at least in the American example—the Congress is superior to the president, because only it can enact trade agreements into law, and it can hold the president accountable by withdrawing his conditional authority to engage in trade negotiations. This is a ubiquitous fact of delegation and accountability relationships: they are vertical relations between subordinate agents and superior principals.

Accountability in Hierarchical vs. Horizontal Relationships

Government, in theory, is an agent of the citizenry in democracies. If this is not the case, then, simply put, the government is not democratic. Of course, under representative democracy, the citizens do not exercise direct authority, rather they delegate it. Alternative forms of constitutional design establish different relations of delegation and accountability as means to further the interests and rights of citizens. There are two basic ways in which constitutions attempt to ensure that the basic rights and interests of citizens as the ultimate principal are not harmed. One is through nested hierarchies, which are typical of parliamentary design. The other is through horizontal exchange, common in presidential systems.6

Nested Hierarchies Parliamentary systems in their purest form consist of a single chain of nested principal–agent relationships. As depicted in stylized form in Figure 4.1, parliamentary democracy entails each entity as a single agent of its immediately (p.86) superior principal.7 Voters make only one voting choice: they select a candidate (or list of candidates) to represent them in parliament.8 Parties serve as a screening mechanism for voters, enabling them to select legislators who will in turn select cabinet ministers who share their policy preferences. Voters hold legislators accountable through the shadow of future elections and with the assistance of ‘fire alarms’, i.e. third-party provision of information that assists voter-principals in monitoring their politician-agents (McCubbins and Schwartz 1984). Opposition parties and the media provide these fire alarms for voters by publicizing alleged misdeeds by politicians.

The executive enjoys no constitutional independence from the legislature and no direct connection to the electorate. It is instead a pure agent of the parliament, and accountable to that majority in the most simple and direct way: subject to being ousted at any time by a vote-of-no-confidence. The cabinet is also accountable through oversight committees in parliament or through other institutions such as the ‘question period’ used in the British House of Commons. The bureaucracy is an agent of the cabinet in that, even if individual bureaucrats enjoy civil service protection, they must carry out the legislation and ministerial orders by which their political principals delegate tasks to them. The cabinet can sanction bureaucrats through its control over budgets or, ultimately, the structure of the agencies and the civil service system itself (Moe and Caldwell 1994). In the sense described here, a parliamentary system is a nested hierarchy in that each link in the chain is uniquely accountable to its principal.9

Horizontal Exchange In contrast to the single chain of nested hierarchies found in pure parliamentary systems, presidential (p.87)

The Accountability Deficit in Latin America

Fig. 4.1. Relationships of delegation and accountability: nested hierarchies in a parliamentary system. (Arrows indicate the direction of a delegation relationship, running from principal to agent. Accountability relationships run in the opposite direction, with agent accountable to principal.)

systems establish the executive and legislature as separate and independent agents of the electorate. The president and legislature then transact with one another in a series of exchanges to produce policy and, in theory at least, prevent a ‘tyranny of the majority’ from developing. Instead of being accountable to the legislative majority through votes of no confidence, the executive has a fixed term and serves as a check on the ambitions of the legislative majority. For simplicity, Figure 4.2 depicts a unicameral congress, but in fact most presidential systems are bicameral, (p.88)
The Accountability Deficit in Latin America

Fig. 4.2. Relationships of delegation and accountability and horizontal exchange in a presidential system. (Arrows indicate the direction of a delegation relationship, running from principal to agent. Accountability relationships run in the opposite direction, with agent accountable to principal; double-headed arrows indicate horizontal exchange.)

meaning voters have three elected agents. Policy in this system is a product of a series of exchanges (or transactions) between elected agents, who must negotiate a set of instructions for their shared bureaucratic agents. These exchanges frequently spill over into the public arena, as each agent of the electorate seeks to promote its own preferred policy outcomes and blame the other for the failure to enact its own conception of popular preferences. This spillover of disagreements arising out of horizontal exchange serves a fire-alarm function in the sense of providing information for voters to use in assessing the performance of their agents, and voters can use this information in subsequent elections. Because of the separation between executive and legislature, each has an incentive to point out misdeeds by the other, which in theory at least, helps reveal information about scandals or other breaches of faith. In all presidential constitutions, the Congress has the authority to demand information and hold hearings into alleged executive-branch wrongdoing. The information produced by public interbranch disputes thus augments that provided by opposition parties and the media and implies that presidential systems potentially provide (p.89) more information for voter-principals to use in monitoring their agents than do parliamentary systems.10

Bureaucrats in presidential systems, then, are accountable to two principals (three in a bicameral system). They are accountable in that their political principals define through a ‘contract’ the process by which decisions are to be reached and what interests are to be represented in an agency's decision-making process. Administrative procedures ensure that affected interests have a right to participate in rule-making (McCubbins, Noll, and Weingast 1987, 1989). Moreover, legislators build into the enabling legislation of each bureaucratic agency fire alarms whereby the constituents that they seek to have represented in the decision-making process can inform legislators of bureaucratic transgressions. We know less about how this form of decentralized monitoring and accountability of agents works in Latin America than we do about the United States, but growing evidence suggests that the basic logic holds.11

Courts in Hierarchical vs. Transactional Systems

In a pure hierarchical political system, the courts are essentially another part of the bureaucracy (Ramseyer and Rosenbluth 1993), responsible for applying the laws enacted by parliament on behalf of the citizenry, and not empowered to overturn laws. By contrast, in transactional presidential systems, where there is no notion of parliamentary sovereignty, courts typically have authority that overlaps with the elected bodies and may even overturn acts of the elected bodies on constitutional grounds. Increasingly, even in parliamentary systems, constitutional courts or other bodies are being endowed with sufficient independence to serve as checks on the parliamentary majority (Lijphart 1999; Stone 1992), though the origin of constitutional arbiters almost always is based in appointment for relatively short terms by elected agents of the electorate. That is, they typically undercut the sovereignty (or tyranny, if one (p.90) prefers) of the legislative majority, but do not enjoy much separation of origin and survival from legislative politicians.

Much of the action of courts in democracies concerns applying and enforcing the law. Even without a right to declare laws themselves unconstitutional, courts would be expected to be more active in presidential systems than in parliamentary, because bureaucrats serve multiple principals (Moe and Caldwell 1994; McCubbins, Noll and Weingast 1987, 1989). Because of the transactional nature of law-making in presidential systems, there is more potential for conflicting interpretations of legislative intent. There are also the built-in fire-alarm procedures referred to above, which often explicitly allow affected parties to sue agencies for alleged failure to conform properly to procedures or to allow necessary evidence into their decision-making process. Courts can be a tool for ensuring the vertical accountability of bureaucracies to their political principals, especially legislators.12 This may even be the most fundamental function of courts in modern democracies. This is why—as we shall develop below—it is so important that legislators be real partners with the executive branch in policy-making, for if they are not, they are far less likely to have an interest in independent courts as a check on the executive and its bureaucracy.

In presidential systems—and increasingly in parliamentary systems, too—judicial bodies are taking on more and more the role of a check on the legislative process itself, in addition to ensuring the proper application of laws duly enacted by legislators. Courts that have the authority to veto legislative acts are thus another actor in the process of horizontal exchange. Courts are independent to the extent that they are not accountable to the political bodies. They are engaged in horizontal exchange to the extent that they share powers in some areas with the political bodies—or with other independent agencies.

Superintendence Agencies

O'Donnell's definition of horizontal accountability appears to be directed less at the legislative branch or the judiciary and more at the assemblage of so-called autonomous agencies of government (p.91) established in numerous constitutions in Latin America. These include the various components, aside from the judiciary, of the justice system: the Contraloría, Fiscalía, Defensoría, and other entities charged with watching over the actions of the executive (and sometimes legislative) branch. Sometimes these entities have sanctioning authority, but often their primary function is to pull a ‘fire alarm’ when they witness misdeeds. For instance, they may publicize misdeeds and perhaps refer them to the courts or to a congressional committee charged with opening an impeachment inquiry. That is, they substitute for the reticence of regular institutions of horizontal exchange and legislative oversight to detect and punish official wrongdoing, for various reasons that are often rooted in a breakdown of vertical accountability institutions between legislators and citizens. In the following section, we explore the complex relationship between vertical accountability, horizontal exchange and superintendence, especially as it pertains to superintendence agencies in presidential systems.

Vertical Accountability, Horizontal Exchange, and Superintendence

Presidential systems entail a mixture of vertical accountability and horizontal exchange. Additionally, Latin American systems are creating more and more superintendence agencies. These oversight entities are part of the horizontal structure of the state to the extent that they have constitutional status alongside the executive, legislative, and judicial branches, but one of their most important functions is to assist vertical accountability by providing information that elected politicians and citizens can use in holding their agents accountable. Some of the oversight agencies also have prosecutorial functions, bringing cases of alleged legal violations before the judiciary or imposing their own sanctions.

The Madisonian principles of constitutional design that modern presidentialism rests upon rely on horizontal exchange between agents with different vertical accountability ties to the citizenry. In Madison's famous phrase, ambition must be pitted against ambition. The notion is that of countervailing incentives, such that legislators and the executive do not collude with one another to the detriment of the principal (citizens). Countervailing incentives between legislators and the executive arise when each represents a different manifestation of the electorate, through the different (p.92) electoral systems or drawing of constituencies. The ambitions of legislators are to a significant degree set by the institutions of vertical accountability between them and their constituents. In this section, we discuss the interplay between vertical accountability and horizontal exchange. We argue that the nature of the vertical accountability relationship between voters and legislators is crucial to the functioning of a system of horizontal exchange. We argue further that when vertical accountability breaks down, such that legislators fail to be good agents of the citizenry, there is a tendency for constitutional reformers to see superintendence agencies as a solution. However, these entities are a poor substitute for a well functioning interplay of vertical accountability and horizontal exchange.

Vertical Accountability vs. Horizontal Exchange Legislators in presidential systems are vertically accountable to citizens just as in parliamentary systems, so it follows that the electoral system that defines the nature of the hierarchy between voter-principals and legislator-agents is a crucial aspect of the design of a presidential system. The fundamental difference between presidential and parliamentary democracies is that the hierarchical connection between voter-principals and the executive authority is not mediated through the legislative majority in a presidential system as it is in a parliamentary system. That is, whereas parliamentary democracies are based principally on nested hierarchies of vertical accountability, presidential systems are built on the interaction of horizontal exchange and vertical accountability. This fundamental distinction has serious consequences for legislative incentives in presidential systems (Shugart and Haggard 2000), and, in turn, for how well horizontal exchange functions to align the incentives of actors in the various branches with the interests of the ultimate principal, the citizenry.

It has been argued in the literature on the United States presidential system that legislators are single-minded in their pursuit of re-election and that they have designed the institutions of the United States Congress in order to facilitate that goal (Mayhew 1974). A large literature has developed that traces the organization of Congress—especially the House of Representatives—as a series of institutions that enable legislators to ‘claim credit’ for particularistic services to their constituents (e.g., Weingast 1984; Shepsle 1986, Weingast and Marshall 1988). It is not that legislators in parliamentary systems are not as interested in furthering their own political careers; rather what distinguishes parliamentary (p.93) legislators is that the parliamentary majority is collectively responsible for governing. As a result, the pursuit of individual political careers takes place within a framework of much stronger parties than we find in the United States, because such parties are needed to bind the nested hierarchies of parliamentarism to one another. The absence of (or reduced role for) collective responsibility in presidential systems means that it is harder to achieve vertical accountability of the entire policy-making apparatus to the broad preferences of the citizens. Hence mechanisms of oversight and sanctions are handed off to non-elected agencies in an effort to rein in the citizens' agents and reduce shirking.13

In a parliamentary system, one of the principal commitments legislative candidates have to offer their voters is their engagement with a collective entity, a party, that seeks to claim executive authority, or a share of it (Epstein 1967; Cox 1987; Shugart and Carey 1992; Palmer 1995; Moe and Caldwell 1994; Strøm 2000). This collective commitment does not exist in presidential systems because the electoral connection between voters and the executive does not run through legislators, but is direct, via presidential elections. Because legislators are bypassed in the connection between voters and the executive, voters tend to demand different things from legislative candidates than they demand from the executive. This phenomenon has been noted in the United States literature (Jacobson 1990; Moe 1990; Moe and Caldwell 1994). Presidents are held accountable for overall management of the economy—and, given the United States' status as a great power, for foreign affairs. Legislators, on the other hand, are held accountable to a large degree for distributive policies, and are also expected to deliver pork-barrel favours to their districts.

If this distinction between what legislative candidates and presidential candidates offer their voter-principals is significant in the United States, it is even more so in other presidential systems. Latin American electoral systems tend to one or the other extreme, compared to the participatory nomination procedures and single-seat districts of the United States (Shugart 2001). They tend either to emphasize party even less than in the United States, as a (p.94) result of intraparty competition—as in Brazil, Colombia, and Peru—or to centralize party control at the expense of individual legislators, as a result of closed party-list systems—as in Argentina, Mexico, and Venezuela. Thus at the ‘weak-party’ extreme, legislators have even less incentive to offer voters a commitment to a party and its national-policy priorities than in the United States, while at the ‘strong-party’ extreme, legislators are not individually accountable to their constituents at all.

If legislators in presidential systems lack a balance between collective accountability to parties and individual accountability to constituents, a crucial link in the interplay between vertical accountability and horizontal exchange is lacking. If legislators are either uninterested in national policy (excessively weak parties) or unaccountable to voters (excessively strong parties) they have little interest in exercising oversight and ‘political control’ over the executive. That is, they are neither partners in the crafting of national policy, nor engaged in ongoing monitoring of the executive branch's policy-making process. Instead, much of the policy relationship between legislator-principals and the executive as an agent to carry out legislation will look more like abdication than delegation, and the dominant feature of interbranch transactions will not be exchanges over broad policy but exchanges of votes for patronage. It is in this context, then, that the attention of ‘good-government’ reformers turns to the creation of more and more ‘independent’ superintendence agencies to augment or replace the oversight and sanctioning power that should be provided by legislators, through horizontal exchange, and by voters, through vertical accountability of law-makers.

Horizontal exchange involves the existence of independent agencies that check one another. As we have argued, horizontal exchange is not a form of accountability because the very notion of independent branches is that they are not accountable to one another. Indeed, their independence and lack of mutual accountability, along with the overlap of functions, is what allows them to function as checks and balances. No branch can effectively serve as a check on the other if its authority is derived from serving as the agent of the branch to be overseen. The only senses in which relationships between separate branches are ones of accountability, are first, in the assignment of discrete tasks (such as the trade-negotiation process referred to earlier), and, second, impeachment. When one branch initiates impeachment over another, the officials subject to possible impeachment are accountable to another branch—always the legislature, though the legislative authority of (p.95) impeachment may be shared with other agencies in some systems. The evocation of impeachment proceedings, then, is the conversion of a constitutionally horizontal relationship between branches into a legal relationship of accountability in which, typically, legislators intercede between the impeached official and the ultimate principal, the voters. Thus relations between branches are horizontal, and they are not relations of accountability. Impeachment is a relationship of accountability, but in being so, it ceases to be horizontal. Having discussed the distinction between horizontal exchange and vertical accountability in greater depth, we now seek to further define the distinction between horizontal exchange and superintendence.

Horizontal Exchange vs. Superintendence A necessary condition for effective horizontal exchange is overlapping functions between branches. In order for horizontal exchange to take place, each branch that is party to the exchange must have something to offer the other in a transaction. For instance, in the standard legislative process of a presidential system, the Congress has the authority to pass—or not pass—legislation desired by the president. The president, in turn, typically has the authority to veto legislation passed by the Congress, and also has some discretion over the enforcement of enacted legislation. Thus each of these two branches has something the other side wants. The president wants votes in Congress, and the Congress wants the president's assent to legislation and his commitment to enforce it vigorously. Such is the essence of exchange: while the branches are independent in the sense that they have separate bases of authority, they are not fully separate in terms of their functions. They need to cooperate with one another in order to accomplish their tasks. Their functions, in short, overlap.

Overlapping functions also may occur with respect to the nonelected agencies. For instance, where there is a constitutional tribunal empowered to exercise review of the constitutionality of laws, legislators and the executive must engage in exchange with the tribunal in order to enact their preferred legislation. This is especially true in those systems in which a court or tribunal may exercise ‘abstract review’, whereby a law that is challenged on constitutional grounds cannot be promulgated until it is approved by the branch that exercises constitutional control; in such cases, entire laws may be effectively rewritten by the tribunal and legislators often incorporate the tribunal's text to ensure promulgation can proceed the second time around (Stone 1992).

(p.96) Another example of non-elected agencies taking part in horizontal exchange may be found in the judiciary-dominant model of impeachment (Kada n.d.). Under this impeachment process, the legislature is not the only branch that determines whether a case against the president (or other official) may go forward. In such cases the Supreme Court or an independent attorney general or public prosecutor must also give assent, and thus functions overlap between elected and non-elected entities, with each holding some authority that the other needs to pursue its preferences.

An agency may be independent (in the sense of having a reasonably high degree of separation of origin and survival), yet not be equipped to engage in horizontal exchange if its functions do not overlap in some respects with those of other branches and, in particular, of the elected branches. Absent overlap of functions, whereby one agency needs the cooperation of another to perform certain tasks, no horizontal exchange—let alone accountability—exists. For instance, the ombudsman's office in Peru is a constitutionally defined office with some degree of separation of origin, in the form of an extraordinary congressional majority for appointment. Yet the authority delegated to it by the constitution is quite limited. According to Kenney (this volume), the ombudsman can do little more than investigate and publicize violations of rights; even this function is severely limited by the military's legal right to withhold information for ‘security’ reasons. The ombudsman is not engaged in exchanges with any other branches that need the ombudsman's cooperation in order to perform their own tasks.

If the ombudsman is not institutionally equipped to engage in horizontal exchange, this does not mean it is irrelevant. Indeed, in publicizing misdeeds, a credible ombudsman or other independent oversight agency may play a vital role as a fire alarm in vertical accountability. Recall that fire alarms are third-party opportunities to reveal an agent's misdeeds to the agent's principal. If shedding light on abuses of human rights by agents within the executive branch provides information that legislators can use to sanction the executive, or that voters can use in the next election, then the ombudsman has played a role in enhancing vertical accountability. For another example of actions by an independent agency that is not horizontal exchange, suppose a Fiscal exists who can blow the whistle on corruption. If the Fiscal has sufficient independence (from his origin and survival) to reveal the corruption of a legislator, that information may be useful to the legislator's principals, the voters. The extent to which this information leads to punishment of the legislator, then, depends on the extent (p.97) to which the electoral process facilitates the accountability of legislators, as well as the preferences of the voters for non-corrupt legislators. The extent to which credible information about official misdeeds will lead to the ultimate principal—voters—holding the official accountable therefore depends on institutions of vertical accountability. We return to this theme below.

Countervailing Ambitions A condition for horizontal exchange or superintendence to work is countervailing ambitions. Madison's pitting of ambitions against one another assumes that agents have incentives to do different things. For instance, firms often require large expenditures to be approved concurrently by both the CEO and the controller. The CEO is rewarded for increasing profits, but the controller is rewarded for holding down costs. Both are agents of the firm's board (and, through the board, of the stockholders), but they have countervailing ambitions in the sense that each has a different incentive. In political systems founded on a horizontal separation of powers, similarly, separate institutions must have different incentives.

In presidential systems, countervailing ambitions are implied by separate electoral origin of the president and executive. Presidents are elected in a nationwide race for a single office, while legislators are typically elected at least partially in regional districts. Even in those rare cases of single nationwide legislative districts (e.g., Peru and the Colombian senate), the use of a non-majoritarian electoral system implies that the congressional majority is accountable to a different manifestation of the electorate than is the president.14

The greatest threat to the existence of countervailing ambitions is the presence of informal hierarchical relations between formally independent branches. If a hierarchy exists between separately elected agents of the electorate, then separation is replaced by accountability to a single principal. An example occurs under what might be called majoritarian presidentialism. For instance, if the president is the head of the majority party in the sense of nominating the party's legislative candidates and offering them post-legislative ‘pensions’ in the form of government patronage, then the president becomes a principal over legislators. In this scenario—which approximates to that seen in Mexico until recent years (p.98) (Weldon 1997)—legislators and president continue to have formally separate origin and survival in the electoral process, but cease to have countervailing ambitions.15 While full-blown majoritarian presidentialism is rare, approximations to it may be found at various times in many Latin American countries, undercutting both the vertical accountability of politicians to voters and the horizontal exchange that is supposed to keep the elected branches in competition with one another for public support. In such situation, demands for the creation of superintendence institutions are likely to arise.

Even without a majority party headed by the president, the countervailing ambitions on which horizontal exchange depends can be overridden by relations of de facto accountability that make the legislators dependent on the president. Where legislators win office primarily on patronage rather than on policy-based campaigning, they are in a dependent position vis-à-vis the president, whose position at the head of the executive branch gives him vast patronage resources with which to ‘buy’ votes in congress. Although not a formal relationship of accountability, this sort of interbranch bargaining of votes for patronage undermines the countervailing ambitions of the branches by inducing legislators to abdicate the autonomy of their branch and hence its ability to serve as a forum for the reconciliation of policy disagreements. In such situations, which are probably very common in Latin America, corruption is likely to be rife (because legislators' demand for patronage induces them to look the other way, or even be party to, bureaucratic misappropriation of funds, for example). Again, demands for the creation of independent institutions of superintendence are likely to arise.

It follows from this discussion that either situation—the president as the head of a majority party or parties so weak that patronage pervades legislators' election campaigns—undermines countervailing ambitions. Both situations place the executive in a de facto superior position and short-circuit horizontal exchange. It follows further, then, that situations of moderately strong parties, none of which is subordinate to the president, are the most promising (p.99) means of ensuring countervailing ambitions. Parties are the principal means of ensuring collective accountability of policy-makers to voters, so their virtual absence as coordinators of the legislative process makes legislators highly dependent upon executive patronage. Parties and legislators must retain sufficient independence from the president that they can represent the preferences of their voter-principals.

In courts and superintendence agencies, the same fundamental limitations on countervailing ambitions apply. For example, when the same unified party controls both elected branches, justices are unlikely to be appointed who are independent of politicians even if both branches must participate in their appointment (see Magaloni, this volume). If the overwhelming majority of legislators is primarily interested in patronage, they will have little incentive to select officials for the non-elected agencies who will exercise effective control over the source of patronage, the executive, even if the selection process requires a supermajority. Thus the independence of the non-elected agencies is largely dependent upon the incentives of legislators who play a part in staffing them. If the vertical delegation from voters to legislators is not designed properly to reflect the collective wishes of the electorate, then oversight is unlikely to work and politicians may be able to collude to oppress citizens. We take up the issue of independence across several non-elected superintendence agencies in the following section.

Degrees of Independence of Non-elected Agencies in Latin America

We posit that the proliferation of superintendence agencies across the region is indicative of a breakdown in the interplay of the vertical and horizontal dimensions of the state that we have identified up to now. A crucial question remains as to whether agencies of superintendence are endowed with genuine independence vis-à-vis the elected branches so that they can compensate for the malfunctioning of countervailing ambitions between the elected branches. As we shall see, the picture is mixed, but a great many are dependent on elected branches for their origin and survival. In the following pages, we briefly review the provisions for separate origin and survival as well as some of the key overlapping functions in Latin America's non-elected agencies charged with oversight or control of politicians. We will consider the following (p.100) classes of non-elected agencies: supreme courts, constitutional tribunals, attorneys general (state legal counsel), prosecutors general (public ministries), public defenders (human rights ombudsmen), and controllers general. Rather than explore the entire gamut of non-elected agencies, we have chosen to focus on the agencies of the ‘justice system’ that are found in many constitutions of Latin America because they are part of the ongoing process of maintaining accountability and ensuring compliance with the constitution and law.16

Functions of Non-elected Agencies Justice systems across the region have grown increasingly complex, with distinct agencies performing myriad functions. For instance, supreme courts or constitutional tribunals are engaged in horizontal exchange to the extent that they are empowered to overturn legislation that they find unconstitutional. This role most closely conforms to our definition of horizontal exchange when it is exercised in the abstract, that is, before a bill may be promulgated.17 Of course, the most basic function of the Supreme Court or constitutional tribunal is to uphold the rule of law. Many of the oversight and exchange functions of other non-elected agencies depend on the willingness of the high courts to sanction violators of constitutional rights. Thus if the highest courts are not geniunely independent, it is likely that other non-elected agencies will fail to perform their duties.

Attorneys and prosecutors general are charged with ensuring obedience of the law. The precise terms used vary from one constitution to another, and often posts with different functions go by the same name (e.g., Procurador General or Fiscal General) in different countries. For present purposes we have made a distinction between those bodies whose primary role is to oversee the enforcement of criminal law and/or to represent the government in judicial proceedings (which we call the attorney general), and those whose role extends to ensuring public officials' compliance with the law and investigating corruption (which we call the prosecutor general). Typically, these two functions are performed by separate entities.18 The attorney general or state legal counsel is sometimes (p.101) a direct executive appointee (especially where the official's role is to serve as the government's legal advocate), while the prosecutor general typically enjoys greater independence from the president and the legislative majority. Prosecutors general typically serve the role of the citizenry's advocate against official wrongdoing, making them clearly agents of superintendence. The powers of this organ are sometimes quite sweeping, as in Brazil (see Sadek and Cavalcanti, this volume). Powers can even extend to the suspension of officials from office for reasons of corruption, as occurred in Colombia in April 2000, when the Procurador General, for the first time in the nation's history, suspended two sitting members of the Congress for alleged misappropriation of funds. In this sense, they are adjuncts to vertical accountability, and their presence—and, often, considerable independence—can be seen as a product of the relative difficulty of holding officials accountable in presidential systems through ordinary electoral and legislative institutions, due to fixed terms, separated powers, and faulty electoral laws. Ideally, the accountability functions of legislative hearings (oversight of the executive and internal policing of the legislature's own members) and elections would be sufficient to deter or punish official malfeasance. However, where standard processes of vertical accountability break down, constitution-makers seek recourse in independent agencies to carry out these functions.19

Public defenders for human rights—sometimes called human rights ombudsmen—play an important role in providing fire-alarm oversight. Not all Latin American constitutions make provisions for human rights defenders. Those that do often place the defender within the public ministry (under the prosecutor general), but nonetheless establish a separate organ, sometimes called the Defensoría, charged with investigating human rights abuses. Most (p.102) human rights defenders lack enforcement or sanctioning powers. Their duties are mainly confined to publicizing abuses, thereby serving a fire-alarm function, and to referring abuses to the prosecutor for possible judicial action.

Finally, controllers general are authorized to audit officials and governmental agencies and participate in the formulation and execution of the national budget. Therefore, they have a series of tools at their disposal that may facilitate holding other officials, primarily in the executive branch, accountable in the sense of ensuring compliance with budgets and other laws passed by the Congress. Controllers engage not in fire-alarm oversight, but in the more active and direct form of oversight dubbed ‘police-patrol’.

Assessing the Independence of Agencies of the Justice System Now we turn our attention to the degrees of formal independence enjoyed by the agencies of the justice system. We assess independence in two dimensions, the appointment process, and tenure in office. The appointment process is considered to offer minimal independence if it is dominated by legislators, although appointment by an extraordinary majority of legislators affords somewhat greater separation than selection by a mere majority. Mixed appointment processes offer somewhat greater independence—less so if appointment involves two elected branches (president and Congress) than if it involves politicians and non-elected actors (courts or councils of civic groups). Some constitutions employ mixed procedures involving both judicial councils and politicians. For example, the Venezuelan constitution establishes a council called the Poder Ciudadano20 that nominates candidates to the very same positions that comprise the council, and the nominees are confirmed by the Congress. Also in the mixed category are some judicial councils that actually contain a majority of politicians among their membership (as in the Dominican Republic) or leave it entirely to the Congress to decide how the council shall be constituted (as in Ecuador). Because of the opportunity for direct political influence in these councils, there is somewhat less separation of origin than at first meets the eye.

Considerably greater independence results from a judicial-dominant process, leaving politicians out of the selection of justices or superintendents. For instance, in Colombia and Ecuador the Supreme Court itself sits as a judicial council for the purposes of (p.103) filling vacancies in the court. Chile uses a form of this procedure, in which politicians have a veto over nominees originating from within the Supreme Court itself. Judicial-dominant procedures establish a self-perpetuating court that is likely to be quite independent of politicians. In fact, one could argue that it goes too far, tending to create a court that is very homogeneous and potentially cut off from the citizenry it ultimately is supposed to serve (Correa Sutil 1993).

The greatest independence is afforded by the still-rare civil-society-dominant process. When civil society dominates the appointment process, politicians are left formally out of the selection entirely, and the process is controlled by groups such as bar associations or a commission of academics. Of course, if these civil society groups themselves are entwined with politicians through patron–client ties, there may be less real independence than meets the eye; nonetheless, this mode of appointment offers the greatest potential for making courts and superintendence agencies true agents of civil society rather than of politicians. Peru is the only Latin American country that currently uses this process (for appointing supreme court justices).21

The second dimension of independence is the tenure in office of the officials that head a non-elected agency. We assess this dimension by means of a ‘term ratio’, which is the ratio of the official's term to the term of the elected branch that is involved in appointing. Life terms are coded as 20 years (thereby producing a high score for term ratio). Where there are two elected branches (or two legislative chambers) involved in appointment, the one with the shorter term is used.22 Thus, for example, the United States Supreme Court's term ratio would be 20/4 = 5, because justices are appointed for life terms by a president who has a four-year term. The Senate, which must confirm appointments, has a longer term, at six years, and hence is not included in the calculation (though its involvement shows up on the vertical axis). In cases where politically insulated councils alone appoint officials to life terms (including where Supreme Court justices select their own new colleagues), the term ratio is shown as 5.00, equivalent to the (p.104) maximum found in Latin America. These calculations of term ratios are based on an assumption of fixed terms that can be shortened only through extraordinary impeachment procedures. However, in some cases, nominally long terms can be shortened relatively easily, thereby vitiating the supposedly fixed term. Thus, in instances in which appointed officials can be dismissed by a legislative majority, the term is considered effectively zero, regardless of the ratio that would be obtained by considering only the appointment side of the process.23

Term ratios of 1.00 or less imply essentially no independence, because each legislature (or president) has an opportunity to replace the official. Term ratios from 1.01 to 1.49 provide only slightly more independence, because every other legislature (or president) is likely to have a chance to replace or reconfirm the official. Longer fixed terms increase the opportunities for independence by an official of a non-elected agency by enhancing the attractiveness of the post as a career option and by making more remote the prospect of punishment by elected officials.

Figure 4.3 shows the degree of independence of non-elected agencies charged with checking or overseeing the elected branches. Basic data on the appointment and dismissal processes and term lengths of these entities are given in Tables A4.1 to A4.6 of the appendix. The upper-right region of the figure shows where the criteria for ‘independence’ are met, as opposed to high dependence on the elected branches; movement from any point in the two-dimensional space either upward or to the right indicates an increase in independence. From the abbreviation for each country, dashed lines extend to all of that country's high courts and superintendence agencies, so that one can see at a glance how the overall structure of the justice system in that country affords independence or not.

As can be seen, there is a wide variation in the degree of independence of courts and agencies of superintendence in Latin (p.105)

The Accountability Deficit in Latin America

Fig. 4.3. Independence of high courts and superintendence agencies in Latin America

(Source: Appendix Tables.)

(p.106) America. Superintendence agencies—all the entities shown in Figure 4.3 except for supreme courts and constitutional tribunals—are in all but a very few cases located in the region of the figure that implies limited potential for independent action. Notable exceptions are the prosecutor general of the Dominican Republic and the controller general of Chile, which have life terms. Even several supreme courts are located out of the area of the figure that depicts relatively independent institutions. In a few cases, congressional majorities can remove members of the court; this is true even in Peru despite an appointment process that appears to give the Supreme Court maximum independence.24

Constitutional tribunals, where they exist, almost always have less independence than the respective country's Supreme Court. This is consistent with their more ‘political’ function in the sense that one of their primary roles is to rule on the constitutionality of specific laws. Not surprisingly, the elected legislators or constituent assembly delegates who design constitutions have preferred to keep a body charged with reviewing the admissibility of their own laws on a somewhat shorter leash.25 For advocates of legislative supremacy, this is a good thing, while for those who mistrust legislative bodies, this would be a bad thing. The perils of too little independence were demonstrated clearly in the Peruvian case. When President Alberto Fujimori sought to overturn a ruling of the Tribunal concerning his eligibility to run for a third term, he was able, through the atomized legislative majority that was dependent on his patronage, to have the Tribunal magistrates impeached on trumped-up charges. Figure 4.3 shows why this was possible within the parameters of Peru's constitution: a simple majority of Peru's unicameral congress is sufficient to impeach and remove a judge from office, meaning effectively the Peruvian Supreme Court and constitutional tribunal magistrates do not have fixed terms at all.

Prosecutors general vary greatly in the independence they are accorded. Only the Dominican Republic provides its public prosecutor with a very high degree of independence—appointment by a judicial council for a life term—although politicians comprise a (p.107) majority of the council. Also quite independent (formally) is the prosecutor general of Venezuela. On the other hand, those of Ecuador, Guatemala, and Peru, have very independent appointment processes, but lack the security of a fixed term. Strikingly, one of the least independent in formal terms is that of Brazil, where the public prosecutor is appointed by the president, with consent of the senate, for a term of only two years. Thus the considerable behavioral independence identified by Sadek (this volume) does not rest on constitutional provisions for separate origin and survival. The constitution does mandate that the officers within the public prosecution, and the public prosecutor who heads the entity, come from an internal career path, thereby ensuring a high degree of professionalism. Further, the way in which the federal public prosecution is integrated with the decentralized prosecutorial bodies of the states in Brazil's federal system reduces the potential for political influence at the federal level (see Sadek and Cavalcanti, this volume). Nonetheless, the appointment process and exceedingly short term work against the independence of the institution, and thus the relative freedom of the public prosecution from political influence could vary with the commitment to prosecutorial independence of specific federal administrations.

Public defenders or ombudsmen for human rights are generally quite dependent upon Congress for both appointment and survival in office. Among the very few exceptions to this generalization are those of Guatemala and Venezuela, though even in these cases the ombudsman has a short term. In most other cases, Congress is solely charged with the election and dismissal of individuals to this office, although interbranch consent is required for appointment in Colombia.

There are many examples of legislative dominance in the selection process of the controller general, though only in Mexico is the controller a simple agent of the legislative majority. In fact, most of these entities are located in the lower-left area of the figure where independence is at its lowest. There are some exceptions, however. The most notable is Chile, where the controller, although appointed by the consent of the president and senate, is granted a life term. In Argentina, the controller is dependent on politicians, but not those in the majority; the largest opposition party selects the controller. In Costa Rica, Colombia, and Guatemala the controllers also enjoy considerable formal independence, although through different means: selection from a slate prepared by the high courts in Colombia; the involvement of an apolitical screening committee in Guatemala; a term twice that of the legislature's in Costa Rica.

(p.108) Thus the record with respect to independent agencies set up to monitor or check the activities of elected branches (and their subordinates in their respective bureaucracies) is mixed. Some non-elected agencies in Latin America enjoy substantial independence. New or amended constitutions in Argentina, the Dominican Republic, and Mexico have greatly enhanced the independence of their supreme courts (on Mexico, see Magaloni, this volume). On the other hand, the lower-left area of Figure 4.3, where independence is at its lowest, shows the greatest cluster of non-elected justice-system agencies in Latin America. This region includes mainly human rights defenders (who are relatively powerless anyway) and controllers (whose typical relative dependence on the legislature is consistent with the notion of congress as having the power of the purse26). Where superintendence is left to entities that are dependent on political branches for their origin and survival, then creating more and more of them can be only a poor substitute for establishing legislative accountability to the citizenry. That is, if legislators do not care about (and perhaps even benefit from) executive corruption or are not motivated by their own accountability to the voters to rein in human rights abuses, then it is probably too much to ask of officials that are accountable to them to do so. Thus, for the remainder of this paper, we return to the prospects for electoral reforms that would increase the chances that legislators see their role as representing the values and preferences of the broad national citizenry.

Improving Vertical Accountability: Candidate Selection and Electoral Rules

Thus far, we have shown that presidential systems rely on horizontal exchange between elected branches to prevent oppression of citizens and to reveal information about wrongdoing in one branch or the other. We have noted that horizontal exchange between elected branches often fails, and that the separation of powers (p.109) itself undercuts vertical accountability, because the assigning of blame and meting out of electoral punishment is more difficult than in the nested hierarchy of a parliamentary system. Both vertical accountability and horizontal exchange are prone to break down, because executives and legislators might collude with one another, as when legislators seek patronage and thus have reduced incentive to be vigilant with respect to the executive branch. In this context, reformers in presidential systems often turn to ‘independent’ agencies of superintendence in an effort to generate oversight and compliance that is lacking in the electoral and legislative institutions. However, as we have seen, only some of the superintendence institutions are in fact really independent, thus completing a vicious circle: superintendence agencies are created because politicians may fail in their collective duty to provide good government, but politicians usually appoint or confirm (and sometimes may dismiss) their own overseers.

In this section, we briefly review some of the critical problems of vertical accountability in Latin America, and suggest solutions to them. We argue that much of the problem of legislators' shirking their duties to provide collective oversight of the executive is rooted in party and electoral systems that fail to connect legislators to the interests of the broad citizenry. This failure comes in two stages of the process of choosing legislators, the nomination process and the electoral rules. We review each stage of the process in turn, but first we briefly consider what would be the ideal system of vertical accountability, keeping in mind its limitations under a system of separated powers.

Effective Vertical Accountability As we introduced the concept of accountability above, we noted that achieving both effective legislative oversight of the executive and effective non-elected agencies requires what we might call ‘getting vertical accountability right’. The apparent dearth of what O'Donnell and others call ‘horizontal accountability’ is largely a malfunctioning of vertical accountability. We do not claim that strengthening vertical accountability is a sufficient condition for redressing the accountability deficit, but we do claim that it is a necessary one. As we have already mentioned, the checks and balances required to enhance accountability only emerge through purposive action by elected officials. With respect to the creation or maintenance of superintendence agencies, it is evident that their very existence and capacity to check the powers of elected officials are dependent upon the roles/resources assigned to them by elected officials.

(p.110) As we noted above, the design of presidential systems requires the elected executive and legislative branches to work together to produce policy outputs. It further assumes that exchanges between the executive, legislative, and judicial branches will prevent oppression of citizens, because the checks and balances among the several branches will prevent the tyranny of concentrated authority. Put simply, horizontal exchanges (among separate branches) promote vertical accountability (of the entire governance system to the citizenry). However, this interplay is prone to failure if the crucial link between voters and their elected legislators is not functioning well. If parties are excessively top-heavy, dominated by national leaders, then legislators will have little incentive to articulate the interests of the voters whom they supposedly represent, and hence are unlikely to organize the legislature in such a way as to serve as an independent actor in the policy-making process. On the other hand, if parties are excessively weak, lacking discipline over individual members, then the legislature will be unable to engage in the collective action needed to serve as a counterweight to the executive. Again, in this situation, legislators are unlikely to invest in making the legislature an independent policy-making actor. While superintendence institutions added onto the horizontal dimension of the democratic state may assist in holding the overall governance structure accountable, as we saw in the last section, legislators are usually deeply involved in selecting those very same superintendents. If legislators lack the incentive to exchange with the executive over policies preferred by their constituents, and are instead motivated primarily by patronage, then superintendence institutions may fail to perform their roles. Thus, a prominent task for political reform in Latin America must be to redesign incentive structures that maximize the potential for effective vertical accountability, so that legislators advance the broad interests of their citizen-principals. We do not argue that such redesign SO would be a panacea, or that the various superintendence agencies could then be abolished, but we do argue that fundamental party and electoral reform is a necessary condition for improving the quality of democracy in the region. We now take up candidate selection (an internal party process) and electoral rules, in turn.

Nomination Processes The candidate selection process is chronologically the first opportunity to establish the relations of vertical accountability. It precedes the general election and, if poorly designed, can prevent what might otherwise be appropriate electoral (p.111) arrangements from establishing a link of accountability between the electorate and legislators. For vertical accountability to work legislators must feel both the need to serve a particular district/constituency but to do so as members of a partisan delegation. Candidate selection processes that are overly decentralized, allowing a group of partisans who constitute only a small subset of the partisans in the district to choose nominees, will err in the direction of emphasizing district over party (constituency service of narrow interests). Processes that are highly centralized, allowing a group outside of the district to choose nominees (including leaving candidate selection in the hands of the party elite), will err in the direction of emphasizing party over district.

A common shortcoming of candidate selection processes in Latin America is excessive centralization—leading to legislators who are responsive to party elites rather than voters. A prominent example was the Venezuelan system from 1958 till at least the 1990s, in which the central party leadership could replace and reorder candidates submitted by state-level parties. Internally undemocratic parties where national leaders hand-pick legislative candidates eliminate any sense of connection legislators might feel to a particular electoral district. Highly centralized candidate-selection processes increase the prospects that a nominee with no personal history in a district or special knowledge of its characteristics will be imposed upon the district from the outside.

At the other extreme, it is possible to imagine very small portions of the selectorate placing candidates on the ballot who would then be beholden to those very narrow interests rather than the interests of the party as a whole. This, of course, would only lead to a breakdown in the programmatic aspect of vertical accountability if the electoral system did not somehow compensate for these particularistic incentives by assuring that narrowly selected candidates could not be elected without a broader base of support. Electoral systems with very high district magnitudes and highly proportional seat allocation formulas, for example, would not compensate for decentralizing tendencies in the candidate selection process. Candidates would have every incentive to cater to the very defined needs of a particular selectorate (and electorate) rather than thinking of the district as a whole, not to mention the party platform.

In other systems there is virtually no formalized candidate-selection process at all. In such systems the prospects for vertical accountability are determined by the electoral rules alone. Where the electoral rules encourage particularistic, pork-barrel forms of (p.112) representation, the candidate-selection process cannot counteract them. Perhaps the best example is Colombia where party leaders do not exercise any effective control over candidates' use of the party label. These legislators then have incentives to deliver patronage or ‘pork’ to a small subset of their parties' following. Similarly in Brazil, the candidato nato clause, which guarantees incumbents the right to renomination by the same party (Mainwaring 1991), in essence eliminates partisan selection of candidates. Incumbents running for office do not have to cater to broader party interests in order to assure themselves a viable place on the ballot.

Electoral Rules The electoral rules are a critical link in any process of strengthening the accountability of the policy-making process to voter-principals. As we have stressed, the fact that the connection between voters and the executive does not run through legislators in presidential systems (as it does in parliamentary systems) means that legislators have little incentive to take into account the consequences of their actions for the executive's national policy program. Thus, if legislators' exchanges with the executive branch are to take place in a context of bargaining over policy, rather than an exchange of executive-offered patronage for legislators' votes, the electoral system must encourage legislators to emphasize voters' policy preferences. In this section we discuss features of electoral systems that help encourage legislators to be collectively accountable to their constituents over policy.

Electoral rules vary greatly in the degree to which they encourage candidates to cultivate a ‘personal’ or a ‘party’ vote (Carey and Shugart 1995). When the incentive to cultivate a personal vote is very high, as a result of competition for votes within parties, candidates tend not to campaign on broad policy issues. The reason is that in systems of intraparty competition, candidates need to emphasize the areas in which they differ from other candidates of the same party. Emphasizing issues, on the other hand, implies focusing on characteristics that they share with co-partisans.

When the incentive to cultivate a party vote is very high, candidates can run as partisan teams, and the potential to articulate shared themes outlined in a party platform is high. However, there is no guarantee that the partisan team will emphasize policy, because of an incentive-incompatibility problem brought on by institutions that create a high party-vote incentive. Such institutions are those that block intraparty competition by requiring voters to give only a party vote. The election of specific candidates (p.113) in such closed-list systems is a product exclusively of the nomination and ranking of candidates on the ballot by the party. Thus the candidate-selection process is crucial within closed lists. If that selection process is decentralized and democratic, it may encourage candidates to remain close to the interests and policy preferences of rank-and-file party activists and voters, while then allowing them to campaign as policy-motivated teams in the general election. However, if the selection process is highly centralized and not subject to democratic procedures, then candidates will cater not to their supposed constituents, but instead to party leaders. In such highly centralized parties, it is sometimes said, accurately, that there is no accountability of legislators to voters.

Because of the dearth of accountability within parties when the nomination process is centralized and the electoral system is closed list, reformers often turn to various systems of preference voting within lists. In such systems, voters reorder the list, encouraging candidates to be responsive to voters instead of just to party leaders. However, having a preference vote means that candidates of the same party are in competition with one another, and as we noted above, such systems generate a personal vote and encourage attention to ways in which co-partisans differ. The surest way in which they differ is in the blocs of voters to whom they deliver pork-barrel and patronage favours.

Thus there is a very delicate balance in the design of electoral rules. Overemphasis on the incentive to cultivate a personal vote risks undermining parties and making legislators conduits of patronage demands rather than of policy preferences. Overemphasis on the incentive to cultivate a party vote risks cutting legislators off from the preferences of voters and makes them dependent on party leaders, especially if the nomination process itself is centralized. It happens that Latin America is home to electoral systems that are very extreme on the incentive to cultivate a personal or party vote. For instance, Colombia elects its senate in a single nationwide district of 100 seats under a formula employing personal lists. Most of these lists elect only one candidate, and no votes are pooled from one list to others of the same party. Brazil uses an open list with an average district magnitude (number of seats per district) of about nineteen. Uruguay and Peru likewise use systems of intraparty competition in large districts. As noted by Carey and Shugart (1995), the premium placed on cultivating a personal vote increases with magnitude in systems in which voters are casting votes below the party level (for one or more candidates or for one of several lists presented by a given party). On the other (p.114) hand, there are also some very high-magnitude districts in Latin America in countries that use closed lists. Many of the closed-list systems, including those of Costa Rica, the Dominican Republic, and El Salvador, have at least one district with over 20 seats. Under a closed list the incentive for members to cultivate a personal vote decreases as magnitude increases, because the longer the list, the more legislators there are in electorally unassailable list positions. The important decisions concerning members' careers are made at the nomination stage, when their list positions are determined, not at the election.

Given the tendency of many Latin American countries to have systems that generate either a very high or very low incentive to cultivate a personal vote, electoral reforms should centre on making these incentives more moderate. With less extreme incentives, candidates' responsiveness to a national party, on the one hand, and to their own constituents, on the other hand, would be balanced.27 One way of achieving a balance between local and party interests is the increasingly popular ‘mixed-member’ system. In such a system some members are elected in single-seat districts (SSDs) and some from party lists, and voters are generally given both a list vote and a vote for a candidate in their SSD (Shugart and Wattenberg 2001). Mixed-member systems generate countervailing ambitions within the party, in that to do well in the overall system, at least the major parties must present both candidates who are attractive to local constituencies and a party label that can attract list votes. Studies of mixed-member systems have shown that even members elected from lists tend to be responsive to localities, probably because their parties recognize the importance of maintaining a profile in SSDs that they may not have won in the last election, but may hope to win in the future (Barker et al. 2001; Klingemann and Wessels 2001). However, we would caution that the hopes of reformers who advocate mixed-member systems could be dashed if the nomination process is not opened up. This is especially crucial for the list portion of the mixed-member system, as often some of the list-tier districts have very high magnitudes.

(p.115) Moderate forms of intraparty competition in general elections may also provide a balance of incentives. The key is not to allow the share of the party's vote—or the total vote—needed for election to be so small that simply providing pork or clientelistic services to a very narrow constituency is sufficient to win a seat. For example, the Chilean system is obviously compatible with meaningful party labels despite its provision for candidate preference voting. No doubt the strength of Chilean parties derives in part from their significant connections to social groups (Valenzuela 1978; Scully 1995), but even a party with a strong social identity must develop a strategy for coping with intraparty competition when the electoral system requires it, as it did in Chile. Since the return to democracy in 1989, there is not actual intra-party competition, because the two-seat districts have encouraged alliances that nominate only one (at most) candidate of any one component party in each district. However, from 1958 through 1969, interparty alliances were not common and the district magnitude was greater than two, yet the preference voting did not undermine party labels. The larger the magnitude, the more co-partisans one candidate must face, and thus the smaller is the average number of votes per candidate. Thus the low magnitudes in pre-Pinochet Chile (average five, largest eighteen) imply larger personal-vote constituencies than the very high magnitudes in Brazil (average about nineteen, largest seventy)28 or formerly Peru (single national district of 120 in 1995 and 2000). Systems of intraparty competition can be consistent with the countervailing ambitions of both local personal-vote-seeking and allegiance to a party where the district magnitude is relatively low because it is not possible to win with extremely narrow personal-vote constituencies—party incentives can counter the personal-vote incentives.

No electoral system single-handedly can transform a personalistic system into one where legislators begin campaigning on party labels and policy. Nor can electoral reform assure that a hyper-centralized system is replaced by one in which individual candidates cater to voter policy preferences rather than primarily to party leaders. However, electoral reform can shift the calculus of candidates in the desired direction, by eliminating or reducing incentives to cultivate a personal vote in personalistic systems or adding a constituency feature of decentralized nominations in (p.116) hyper-centralized systems. Where an electoral system provides for countervailing personal-vote and party-vote incentives, the odds are increased that candidates will be both responsive to local interests and articulators of national policy, because such mixed strategies are more likely to pay off electorally.

The Importance of Balance in Achieving Vertical Accountability  The key to improving vertical accountability of legislators to the citizenry is to find moderation or balance between extremes that will eliminate the tension between representing a particular constituency and representing a programmatic party. Legislators subject to incentive structures at either extreme will not have the countervailing ambitions necessary to check and balance presidents. Those facing excessively decentralized candidate-selection processes will be easily bought out by presidents as legislators pursue resources to distribute to their districts. Regarding the case of Peru, Lourdes Flores Nano notes how vulnerable individualistic legislators lacking any party or ideological commitments are to succumbing to executive-branch pressures.29 On the other hand, legislators from excessively centralized parties will abdicate vigilant participation in the policy-making process in favour of slavish loyalty to the party line, as noted for the case of Mexico by Adolfo Aguilar Zinser.30 Under either extreme, then, legislatures become empty shells devoid of interest to those who want to shape policy outcomes. Organized interests turn their attention to the executive and/or national party offices, neither of which is interested in district-level concerns, and thus the collective accountability of legislators for policy concerns—on which effective horizontal exchange depends—is short-circuited.

Candidate-selection processes and electoral rules are stages of the same process, and, as a result, their combination greatly affects the prospects for getting vertical accountability ‘right’. There is as (p.117) yet little research on the combination of these stages, but we can make some preliminary assessments. It would seem that excessive centralization in the candidate-selection stage is likely to swamp decentralizing incentives at the electoral stage. Thus, for example, the adoption of single-seat districts for about half the seats in Venezuela in 1993 showed little effect on the tendencies of legislators to represent district interests, because the nomination process remained as centralized as ever in the then-dominant AD (Kulisheck and Crisp 2001). On the other hand, an overly open nomination process may vitiate the party-centred institutions at the electoral stage. For example, in Ecuador before 1996, most parties required practically no commitment of prospective candidates to a party organization (let alone a policy platform) to get on the party list, so parties were weak, undisciplined, and personalistic in spite of a closed-list system (Conaghan 1995). If getting vertical accountability ‘right’ in the principal–agent relation between voters and politicians means encouraging candidates to represent the collective preferences of their constituents with respect to policy issues, as we have argued, both the candidate-selection and electoral stages of that relationship must be designed well. Thus reformers need to consider both stages carefully.

Relations Among Agents Are a Function of Their Connections to the Principal

Agents cannot hold other agents accountable, only their principals can. Horizontal exchange based on the policy preferences of the electorate, the ultimate principal in a democratic system, will occur only where the vertical connections between the electorate and its multiple agents encourage those elected officials to act upon countervailing ambitions. Presidential systems provide for separate origin and survival and overlapping functions; candidate-selection and electoral processes (are supposed to) assure that incentives for countervailing ambitions over programmatic policy outcomes lead to horizontal exchange among agents. Where horizontal exchange over policy preferences is lacking it is because the vertical links between voters and elected officials are malfunctioning. In this context, there is a tendency to see the superintendence agencies as a panacea, enforcing ‘accountability’ where legislators fail to check the executive and uncover wrongdoing and where voters are incapable of sanctioning their elected agents.

(p.118) The solution to the accountability deficit in Latin America is not the rapid proliferation of new non-elected agencies of superintendence. These agencies may be helpful in deterring or punishing official transgressions against the ultimate principal (the citizenry), but only as an adjunct to vertical accountability. They are unlikely to function well in a context of especially poor accountability of legislators to citizens, especially when legislators are de facto accountable to the executive instead. Assuring that the vertical links between legislators and voters encourage effective horizontal exchange is thus a necessary first step. Once this is accomplished, the creation of new actors may not be necessary, but, if it is, the prospects that they will fulfil the roles for which they are intended increase dramatically when the legislators who participate in their appointment are vertically accountable to citizen interests. In fact, creating additional agencies without repairing the vertical accountability of legislators will raise expectations that have little hope of being fulfilled. Creating a new agency with much fanfare but then failing to endow it with the necessary faculties to fulfil its role only heightens the public's sense of disillusionment, and undermines the integrity and legitimacy of the very democracy-strengthening project to which the authors of new or reformed constitutions have committed so much effort.

(p.119)

Table A4.1. Constitutional provisions for separation of origin and survival: Supreme Courts in Latin America

Country

Nominate

Confirm/Appoint

Term (years)

Accuse/initiate dismissal

Dismiss

Argentina

President

Senate—2/3

Lifetime

Chamber of Deputies (authorizes trial —2/3 vote)

Senate

Bolivia

Council of Judicaturea

Congress—2/3

10

Chamber of Deputies

Senate—2/3

Brazil

President

Senate

Lifetime, after 2 yrs

(Not identified)

Senate—2/3

Chile

Supreme Ct.—slate of 3

President

Lifetime

Chamber of Deputies

Senate—2/3

Colombia

Council of Judicatureb

Supreme Court

8

Chamber of Deputies

Senate—2/3

Costa Rica

Congress

Congress—2/3 vote

8

Supreme Court—2/3 vote

Dominican Republic

National Council of Magistratesc

National Council of Magistrates

Lifetime

Supreme Court

Ecuador

Supreme Court

Supreme Court

Not specified

National Council of the Judicature

El Salvador

Congress—2/3

Congress—2/3

5

Congress—2/3

Guatemala

Postulation Committee—slate of 26 by 2/3 voted

Congress

5

Congress—2/3

Honduras

Congress

Congress

4

Congress

Supreme Court

Mexico

President

Senate—2/3

15

Chamber of Deputies

Senate—2/3 vote

Nicaragua

President or Congress (w/consultation from civic gps.)

Congress—3/5

7

Congress—2/3 vote (removes immunity)

Tried criminally

Panama

President (with VP & ministers)

Congress

10 (every 2 yrs)

Supreme Court

Paraguay

President from list submitted by Council of Magistratese

Senate

5

Chamber of Deputies

Senate—2/3 vote

Peru

National Council of Magistratesf

National Council of Magistrates

Life

Congressional Permanent Committee

Congress

Uruguay

Congress

Congress—2/3 vote

10

Chamber of Deputies

Senate—2/3 vote

Venezuela

Postulation Committee and Poder Ciudadanog

Congress

12

Poder Ciudadano

Congress—2/3 vote

(p.120)

(p.121)

Table A4.2. Constitutional provisions for separation of origin and survival: Constitutional tribunals in Latin America

Country

Nominate

Confirm/Appoint

Term (years)

Accuse/initiate dismissal

Dismiss

Bolivia

Congress—2/3

10

Chamber of Deputies

Senate—2/3

Chile

3 by Sup. Ct., 1 by Pres, 2 by Nat. Sec. Council, 1 by Sen.

8

Chamber of Deputies

Senate—2/3

Colombia

Slate by Pres., Sup. Ct., Council of State.

Senate

8

Chamber of Deputies

Senate—2/3

Ecuador

2 by Pres, 2 by Sup. Ct., 2 by Cong, 1 by local officials, 2 from civil society organizations

Congress

4

Congress

Peru

Congress—2/3

5

Congressional Permanent Committee

Congress

(p.122)

Table A4.3. Constitutional provisions for separation of origin and survival: Attorneys General in Latin America

Country

Nominate

Confirm/Elect

Term (years)

Accuse/initiate dismissal

Dismiss

Colombia (Fiscal General)

President

Supreme Court

4

Chamber of Deputies

Senate (charges of common crimes referred to Supreme Court)

Ecuador

President—slate of 3

Congress

4

Congress

Congress

El Salvador

Congress

Congress

3

Congress—2/3

Honduras

Congress

Congress

4

Congress

Supreme Court trial

Mexico

President

Senate

Not fixed

President may dismiss at will

Panama

President

Congress

10

(Not identified)

Supreme Court trial

Paraguay

President from list submitted by Council of Magistrates

Senate

Not fixed

President

Venezuela

President

Congress

Not specified

President

(p.123)

Table A4.4. Constitutional provisions for separation of origin and survival: Prosecutors General in Latin America

Country

Appoint

Confirm/Elect

Term (years)

Accuse/initiate dismissal

Dismiss

Bolivia

President

President

Not specified

Chamber of Deputies

Senate

Brazil (Procurador- General)

President

Senate

2

President

Senate

Colombia (Procurador General)

President, Supreme Court, and Council of State—slate of 3

Senate

4

Determined by law

Dominican Republic

National Council of Magistrates (see Table A4.1)

National Council of Magistrates

Life

Supreme Court

Ecuador

National Council of Judicature—slate of 3a

Congress

4

Congress

Congress; National Judicature may also dismiss members.

El Salvador

Congress

Congress

3

Congress—2/3

Guatemala

Postulation Committee—slate of 6 by 2/3 voteb

President

4

President removes for ‘just cause’

Honduras (Director General de Probidad)

Congress

Congress

5

Congress

Supreme Court trial

Nicaragua

President or Congress

Congress—3/5

Not specified

Determined by law

Panama

President

Congress

5

Supreme Court

Paraguay

President (Council of magistrates submits slate)

Senate

5

Chamber of Deputies

Senate—2/3 vote

Peru

National Committee of Attorneys General

National Committee of Attorneys General

3

Congressional Permanent Committee

Congress

Venezuela

Postulation Committee submits slatec

Congress—2/3

7

Supreme Court

Congress

(p.124)

(p.125)

Table A4.5. Constitutional provisions for separation of origin and survival: defenders for human rights in Latin America

Country

Nominate

Confirm/Elect

Term (years)

Accuse/initiate dismissal

Dismiss

Argentina

Congress

Congress—2/3 vote

5

Congress—2/3 vote

Bolivia

Congress

Congress—2/3 vote

5

Attorney General requests trial

Supreme Court

Colombia

President—slate of 3

Chamber of Deputies

4

Determined by law

Ecuador

Congress (w/hearing from human rights organizations)a

Congress

5

Congress

Congress; National Judicature may also dismiss members

El Salvador

Congress

Congress—2/3 vote

3

Congress—2/3

Guatemala

Committee of Congressb

Congress

5

Congress—2/3

Nicaragua

Congress

Congress

Determined by law

Congress—2/3 vote

Tried criminally

Paraguay

Senate—slate of 3

Chamber of Deputies

5c

Chamber of Deputies

Senate—2/3 vote

Peru

Congress

Congress—2/3 vote

5

Congressional Permanent Committee

Congress—2/3 vote

Venezuela

Postulation Committeed

Congress—2/3

7

Supreme Court

Congress

(p.126)

Table A4.6. Constitutional provisions for separation of origin and survival: Controllers General in Latin America

Country

Nominate

Confirm/Elect

Term (years)

Accuse/initiate dismissal

Dismissal

Argentina

Opposition party with most seats in congress

Not specified

Not specified

Not specified

Bolivia (Tribunal de Cuentas)

Senate—slate of 3

President

4

Chamber of Deputies

Senate—2/3

Brazil

1/3 chosen by President; 2/3 chosen by congress

Not specified

Determined by law (criminal)

Chile

President

Senate

Life

Chamber of Deputies

Senate—2/3

Colombia

Constitutional court, Sup. Ct, and Council of State—slate of 3

Congress

4

Determined by law

Costa Rica

Congress

Congress

8, 2 yrs after pres term

Chamber dismisses with 2/3 vote

Dominican Republic

President—slate of 3

Senate

4

Chamber of Deputies—3/4

Senate—3/4

Ecuador

Congress—slate of 3

President

4

Congress

Congress

El Salvador (Corte de Cuentas)

Congress

Congress

3

Congress—2/3

Guatemala

Postulation Committee (2/3 vote)a

Congress—2/3 vote

4

Congress (mental incapacity, crime, negligence)

Honduras

Congress

Congress

5

Congress

Supreme Court trial

Mexico

Congress

Congress

Determined by law

Chamber supervises performance. No specific dismissal procedures.

Nicaragua

President or Congress (w/consultation of civic groups)

Congress—3/5

6

Congress—2/3 vote

Tried criminally

Panama

Congress

Congress

6

Supreme Court

Paraguay

Senate—slate of 3

Chamber of Deputies—2/3 vote

5

Chamber of Deputies

Senate—2/3 vote

Peru

Congress

Congress

7

Congressional Permanent Committee

Congress; National Council of Magistrates also oversees/ratifies members every 7 yrs

Uruguay (Tribunal de Cuentas)

Congress

Congress—2/3 vote

5

Congress—2/3 vote

Venezuela

Committee of Postulationb

Congress—2/3

7

Supreme Court

Congress

(p.127)

(p.128) References

Bibliography references:

Barker, F., J. Boston, S. Levine, E. McLeay, and S. Roberts. 2001. ‘An Initial Assessment of the Consequences of MMP in New Zealand’. In Matthew Soberg Shugart and Martin Wattenberg, eds., Mixed-Member Electoral Systems: The Best of Both Worlds? Oxford: Oxford University Press.

Carey, John, and Matthew S. Shugart. 1995. ‘Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas’. Electoral Studies 14, 4: 417–39.

Conaghan, Catherine M. 1995. ‘Politicians Against Parties: Discord and Disconnection in Ecuador's Party System’. In Scott Mainwaring and Timothy Scully, eds., Building Democratic Institutions: Party Systems in Latin America. Stanford, CA: Stanford University Press.

Correa Sutil, Jorge. 1993. ‘The Judiciary and the Political System in Chile: The Dilemmas of Judicial Independence During the Transition to Democracy’. In Irwin P. Stotzky, ed., Transition to Democracy in Latin America: The Role of the Judiciary. Boulder, CO: Westview Press.

Cox, Gary. 1987. The Efficient Secret: The Cabinet and the Development of Political Parties in Victorian England. New York, NY: Cambridge University Press.

Cox, Gary, and Matthew McCubbins. 1993. Legislative Leviathan: Party Government in the House. Berkeley, CA: University of California Press.

Dunn, John. 1999. ‘Situating Democratic Political Accountability’. In Bernard Manin, Adam Przeworski and Susan Stokes, eds., Democracy, Accountability, and Representation, pp. 329–51. New York, NY: Cambridge University Press.

Elster, J. 1999. ‘Accountability in Athenian Politics’. In A. Przeworski, S. Stokes and B. Manin, eds., Democracy, Accountability, and Representation, pp. 253–78. Cambridge: Cambridge University Press.

Epstein, Leon D. 1967. Political Parties in Western Democracies. New York, NY: Praeger.

Flores Nano, Lourdes. 2000. ‘El Caso Peruano’. Paper presented at the conference Institutions, Accountability and Democratic Governance in Latin America, Kellogg Institute for International Studies, University of Notre Dame, May 8–9.

Grofman, Bernard N. 1987. ‘Will the New Real Institutionalism Please Stand Up and Take a Bow?’ Unpublished paper.

—— . 1989. ‘The Federalist Papers and the New Institutionalism: An Overview’. In Bernard Grofman and Donald Wittman, eds., The Federalist Papers and the New Institutionalism. New York, NY: Agathon Press.

Haggard, Stephan, and Mathew D. McCubbins, eds. 2000. Structure and Policy in Presidential Democracies. New York, NY: Cambridge University Press.

(p.129) Jacobson, Gary. 1990. The Electoral Origins of Divided Government: Competition in U.S. House Elections, 1946–1988. Boulder, CO: Westview Press.

Kada, Naoko. n.d. ‘The Politics of Impeachment’. Ph.D. dissertation. University of California, San Diego.

Kiewet, D. Roderick, and Mathew D. McCubbins. 1991. The Logic of Delegation: Congressional Parties and the Appropriations Process. Chicago, IL: University of Chicago Press.

Klingemann, H., and Wessels, B. 2001. ‘The Political Consequences of Germany's Mixed-Member System: Personalization at the Grass Roots?’ In Matthew Soberg Shugart and Martin Wattenberg, eds., Mixed-Member Electoral Systems: The Best of Both Worlds? Oxford: Oxford University Press.

Kulisheck, Michael, and Brian F. Crisp. 2001. ‘The Legislative Consequences of MMP Electoral Rules in Venezuela’. In Matthew Soberg Shugart and Martin Wattenberg, eds., Mixed-Member Electoral Systems: The Best of Both Worlds? Oxford: Oxford University Press.

Laver, Michael and Kenneth Shepsle. 1999. ‘Government Accountability in Parliamentary Democracy ’. In Adam Przeworski, Susan Stokes, and Bernard Manin, eds., Democracy, Accountability and Representation. Cambridge: Cambridge University Press.

Levy, Brian, and Pablo Spiller. 1993. The Institutional Foundations of Regulatory Commitment, Cambridge: Cambridge University Press.

Lijphart, Arend. 1999. Patterns of Democracy: Government Forms and Performance in Thirty-six Countries. New Haven, CT: Yale University Press.

Mainwaring, Scott. 1991. ‘Politicians, Parties, and Electoral Systems: Brazil in Comparative Perspective’. Comparative Politics 24, 1 (October): 21–43.

Mayhew, David. 1974. Congress: The Electoral Connection. New Haven, CT: Yale University Press.

McCubbins, Mathew, Roger G. Noll, and Barry R. Weingast. 1987. ‘Administrative Procedures as Instruments of Political Control’. Journal of Law, Economics, and Organizations 3: 243–77.

—— . 1989. ‘Structure and Process, Politics and Policy: Administrative Arrangements and the Political Control of Agencies’. Virginia Law Review 75: 431–82.

McCubbins, Mathew D., and Thomas Schwartz. 1984. ‘Congressional Oversight Overlooked: Police Patrols vs. Fire Alarms’. American Journal of Political Science 28: 165–79.

Moe, Terry. 1984. ‘The New Economics of Organization’. American Journal of Political Science 28, 4: 739–77.

—— . 1990. ‘Political Institutions: The Neglected Side of the Story’. Journal of Law, Economics, and Organization, Special issue.

Moe, Terry and Michael Caldwell. 1994. ‘The Institutional Foundations of Democratic Government: A Comparison of Presidential and Parliamen (p.130) tary Systems’. Journal of Institutional and Theoretical Economics 150, 1: 171–95.

O'Donnell, Guillermo. 1994. ‘Delegative Democracy’. Journal of Democracy 5, 1: 55–70.

—— . 1999. ‘Horizontal Accountability in New Democracies’. In Andreas Schedler, Larry Diamond, and Marc F. Plattner, eds., The Self Restraining State: Power and Accountability in New Democracies. Boulder, CO: Lynne Rienner Publishers.

Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theory of Groups. Cambridge, MA: Harvard University Press.

Palmer, Matthew. 1995. ‘Toward an Economics of Comparative Political Organization: Examining Ministerial Responsibility’. Journal of Law, Economics and Organization 11, 1:164–88

Ramseyer, J. Mark and Frances McCall Rosenbluth. 1993. Japan's Political Marketplace. Cambridge, MA: Harvard University Press.

Scully, Timothy. 1995. ‘Reconstituting Party Politics in Chile’. In Scott Mainwaring and Timothy Scully, eds., Building Democratic Institutions: Party Systems in Latin America. Stanford, CA: Stanford University Press.

Shepsle, Kenneth A. 1986. ‘Institutional Equilibrium and Equilibrium Institutions’. In Herbert F. Weisberg, ed., Political Science: The Science of Politics. New York, NY: Agathon Press.

Shugart, Matthew Soberg. 2001. ‘Extreme Electoral Systems and the Appeal of the Mixed-Member Alternative’. In Matthew Soberg Shugart and Martin Wattenberg, eds., Mixed-Member Electoral Systems: The Best of Both Worlds? Oxford: Oxford University Press.

Shugart, Matthew Soberg and John Carey. 1992. Presidents and Assemblies: Constitutional Design and Electoral Dynamics. Cambridge: Cambridge University Press.

Shugart, Matthew Soberg and Stephan Haggard. 2000. ‘Institutions and Public Policy in Presidential Systems’. In Stephan Haggard and Mathew D. McCubbins, eds., Structure and Policy in Presidential Democracies. New York, NY: Cambridge University Press.

Shugart, Matthew Soberg and Martin Wattenberg, eds. 2001. Mixed-Member Electoral Systems: The Best of Both Worlds? Oxford: Oxford University Press.

Stone, Alec. 1992. The Birth of Judicial Politics in France: The Constitutional Council in Comparative Perspective. New York, NY: Oxford University Press.

Strøm, Kaare. 2000. ‘Delegation and Accountability in Parliamentary Democracies’. European Journal of Political Research 37, 3 (May): 261–89.

Tsebelis, George and Jeannette Money. 1997. Bicameralism. Cambridge: Cambridge University Press.

Valenzuela, Arturo. 1978. The Breakdown of Democratic Regimes. Baltimore, MD: The Johns Hopkins University Press.

(p.131) Weingast, Barry R. 1984. ‘The Congressional-Bureaucratic System: A Principal–Agent Perspective’. Public Choice 44: 147–92.

Weingast, Barry R. and William Marshall. 1988. ‘The Industrial Organization of Congress’. Journal of Political Economy 96: 132–63.

Weldon, Jeffrey. 1997. ‘Political Sources of Presidencialismo in Mexico’. In Scott Mainwaring and Matthew S. Shugart, eds., Presidentialism and Democracy in Latin America. Cambridge: Cambridge University Press.

Notes:

(a) Elected by Congress by two-thirds vote from among experienced lawyers.

(b) Seven experienced lawyers elected by Congress, six members elected by high courts.

(c) Composed of the President, Public Prosecutor, President of Senate, a senate member, President of Chamber of Deputies, one deputy, President of Supreme Court, and one Supreme Court justice.

(d) Selection body includes President (or Vice-President), President of Senate, President of chamber, President of Supreme Court, one Supreme Court justice.

(e) Consists of one Supreme Court justice chosen by the Court, one member appointed by President, one by the Senate, one by the Chamber of Deputies, two lawyers chosen by law associations, and two law professors chosen by their colleagues.

(f) Selection body includes officials from other branches (Supreme Court justices) and members of civil society (lawyers, university deans).

(g) The Committee on Judicial Postulations is composed of representatives of societal groups connected with the judiciary; the Poder Ciudadano is composed of the National Public Defender of Human Rights, the Controller General, the Public Minister, and the Public Prosecutor.

(a) Composition determined by law.

(b) Composed of representatives for the deans of universities throughout the country, deans of law departments, deans of law/social science departments of each university in the country and an equivalent number of representatives from the National Assembly of Lawyers and Notaries.

(c) Originates in the Poder Ciudadano branch of government (see note g to Table A4.1).

(a) Selection process includes a hearing in which non-governmental actors may participate but do not vote.

(b) A nominating committee composed of representatives from all represented parties presents 3 nominees. Congress makes a final selection.

(c) Concurrent with legislative term.

(d) Originates in the Poder Ciudadano branch of government (see note g to Table A4.1).

(a) Composed of representatives of university deans of accounting and auditing departments throughout the country and an equal number of representatives chosen by the National Assembly of Economists, Accountants, Auditors, and Business Administrators.

(b) Originates in the ‘Poder Ciudadano’ branch of government (see note g to Table A4.1).

(1) Thus, accountability in this sense is not restricted to sanctions imposed for legal violations, but applies to (potential) sanctions for any failure of subordinates to carry out a task in a manner desired by their superiors.

(2) It is worth recalling that judicial review—the authority of the courts to declare laws unconstitutional—is not mentioned in the United States constitution. It was an authority asserted by the Supreme Court in Marbury vs. Madison. More recent presidential (and some parliamentary) constitutions often make such judicial authority explicit, thereby increasing the range of independent actors with which politicians must exchange.

(3) We are ignoring here the possibility of recall elections, a prospect that exists in some states of the United States and in the 2000 Venezuelan constitution.

(4) There are other reasons, which include agents exercising effective authority in politics (Moe 1984) and the principal (i.e., the citizenry) as a large group, facing serious problems of collective action. On the problems of effective monitoring by collective principals and other principal–agent problems more generally, see Kiewiet and McCubbins (1991, Chapter 2). On the problems of collective action, see Olson (1965) and Cox and McCubbins (1993, Chapter 4). Cox and McCubbins specifically focus on the role of political parties in (partially) overcoming the collective action problems of the citizens and legislators.

(5) Wherever presidents are barred from immediate re-election—as in most Latin American countries—the accountability relationship is severely weakened. See Shugart and Carey (1992: 87–91). Accountability of presidents is most effectively exercised when the president is subject to personal accountability as an incumbent running for re-election. However, where parties are important channels of presidential recruitment they may become agents of accountability if the president himself is term-limited. More amorphously, presidents' desire to protect their legacy or to remain active in politics as ‘elder statesmen’ or in lower elected office, may prevent their accountability from being totally severed even where they are ineligible for re-election.

(6) We are speaking here of pure types, for the sake of illustration. Both forms of accountability relationships are employed to some degree in all democracies.

(7) In Figures 4.1 and 4.2, the ‘superior’ principal (the electorate) is shown at the bottom in keeping with common parlance of authority stemming in a democracy ‘from below’. Nonetheless, in principal–agent terms, those below are in fact superior in the sense of assigning conditional authority to their delegated agents.

(8) The example assumes unicameralism as an essential feature of the pure type of parliamentarism. In fact, many parliamentary systems are bicameral, and even weak upper houses may exercise important powers (Tsebelis and Money 1997). However, nearly all parliamentary systems make the cabinet an agent only of the lower house (i.e., the upper house has no right to cast no-confidence votes). Italy is one of the few exceptions to this rule. On the other hand, most presidential democracies have co-equal upper houses.

(9) As Strøm (2000) has noted, there are informal senses in which the cabinet is directly accountable to the electorate, but there is no formal link between them.

(10) The quality and usefulness of that information is a separate matter, and depends to a large degree on the mechanisms of accountability of legislators and presidents to voters. We take up the issue of legislative accountability in detail below. For a parallel argument about the information content of accountability relations in presidential and parliamentary democracies, see Strøm (2000).

(11) See the various chapters in Haggard and McCubbins (2000) and Levy and Spiller (1993).

(12) Because of the single-line hierarchy, agencies in parliamentary systems are more directly accountable to the executive. Thus the opportunities for third-party intervention through recourse to the courts are diminished. As a result, recourse to the courts in parliamentary systems is more likely to take the form of ensuring compliance with an agency decision than overturning it, let alone overturning a law.

(13) It should be noted that in those parliamentary systems in which fragmentation and factionalism lead to an obscuring of collective accountability, superintendence organs often take on an important role as they do in Latin America. An example would be Italy, where the pervasive corruption was finally brought to official light, and sanctions were imposed, through the work of independent magistrates.

(14) Related to this point, the much lower threshold for obtaining a seat, relative to the presidency, means that a Congress member's constituency can be geographically compact, notwithstanding the nationwide district.

(15) It should be noted that this model of majoritarian presidentialism is quite different from the model of majoritarian parliamentarism. In the latter, the executive is the agent of the legislative majority, accountable to it. In majoritarian presidentialism, accountability relations run in the opposite direction—from the chief executive (in his capacity as party leader) to legislators—implying a much greater concentration of authority in the executive branch.

(16) This list is not exhaustive. For example, some constitutions provide for independent entities charged with controlling corruption, while others place this function in the offices of the prosecutor general and/or controller general.

(17) Abstract review is usually performed by constitutional tribunals, but less often by supreme courts, which usually perform judicial review only in concrete cases (i.e., in controversies that arise after promulgation), if at all.

(18) In some cases, these two functions are unified under one attorney general.

(19) In the United States, the creation of the Office of Independent Counsel (OIC) is a similar response to the difficulty of detecting and prosecuting scandals in the separation-of-powers system. Not surprisingly, given that the OIC was created by legislation and not by constitutional amendment, the reach of the OIC did not extend to the legislative branch, as does the reach of some public ministries in Latin America. Wrongdoing by legislators in the United States is handled by Ethics Committees within the Congress itself. The OIC's reputation for being out of control as a result of Kenneth Starr's pursuit of President Bill Clinton led Congress to allow the statute to expire. As a result, the United States, at least for now, is relying on the Department of Justice (headed by an executive cabinet secretary), Congress's own judiciary committees, and the electoral process to keep executive misdeeds in check.

(20) The council is composed of several officials, including the Human Rights Defender and Prosecutor General.

(21) However, as we shall see, the potential independence that this procedure brings to the Supreme Court is vitiated in Peru by the ease with which a congressional majority may impeach justices.

(22) The elected branch with the shorter term is used because the more frequently elected branches turn over relative to non-elected, the more independent the latter will tend to be vis-à-vis the elected officeholders of the day.

(23) The Ecuadorian case is the most difficult to code. No specific term for Supreme Court magistrates is mentioned in the constitution; in fact Art. 202 states that magistrates do not have a fixed term. However, unlike other officials of the Ecuadorian justice system, Supreme Court magistrates are not subject to a ‘political trial’ by the congressional majority. However, apparently they can be disciplined by a National Judicial Council, whose structure and functions are determined by law. Because this provision appears to leave ample room for political influence, we have considered the term of office of Ecuadorian Supreme Court magistrates to be effectively zero.

(24) As noted by Flores Nano (2000), the executive's compliant congressional majority passed legislation that circumvented the formally independent appointment process.

(25) In Ecuador, there is a mix of different entities—some political and some not—charged with nominating magistrates for confirmation by Congress (see Table A4.2). We have coded this as roughly equivalent to nomination of the entire body by a politicized council (4 on the vertical dimension of Figure 4.3).

(26) Of course, as we have argued above, if the legislature itself has little incentive to see that the executive correctly uses funds—perhaps because misappropriated funds provide patronage or personal-enrichment opportunities for legislators—then a controller's office that is an agent of the legislature is doomed to fail. For this reason, Colombia and Guatemala have increased their controller's autonomy, as shown in Figure 4.3. It is also clear from Figure 4.3 why the controller general in Chile is seen as such an important institution in that country.

(27) In parliamentary systems this is not such a delicate balance because the executive itself is collectively accountable to the parliament and therefore legislators themselves are collectively accountable for their position for or against the government and its policies. In presidential systems, the separation of executive and legislative powers implies greater opportunity for legislators to go their own way and duck governing responsibility, if they have electoral incentives to do so.

(28) In Brazil, the share of votes needed to win is made even smaller by a provision that permits parties to nominate one-and-a-half times as many candidates as there are seats in the district.

(29) Comments prepared by Lourdes Flores Nano (2000), a prominent Peruvian legislator and later a presidential candidate, at the conference on which this volume is based.

(30) Comments prepared by Adolfo Aguilar Zinser, an independent senator, at the conference on which this volume is based. Aguilar Zinser further noted that the situation in Mexico is not only one of absolute party control over nominations, but also of a ban on consecutive reelection. With no prospect for congressional careers, legislators have no incentive to invest in the institutional integrity of the legislature, as after three or six years (depending on the chamber) they will move on to another job—typically a patronage position provided by their party or the executive.